[Feature Request] Storage Reduction for id fields

[mgodwan]

Is your feature request related to a problem? Please describe

Today, _id field is:

1. Indexed using the FST data structure
2. Stored using stored field mapper

For time-series data, the generated _id is not used very often as a query term. The _id field generated is optimized for query but by generating it in a way that we have longer common prefix, we may be able to reduce the storage size it takes.

OpenSearch/libs/common/src/main/java/org/opensearch/common/TimeBasedUUIDGenerator.java

Lines 40 to 42 in 71aefa5

	* These are essentially flake ids but we use 6 (not 8) bytes for timestamp, and use 3 (not 2) bytes for sequence number. We also reorder
	* bytes in a way that does not make ids sort in order anymore, but is more friendly to the way that the Lucene terms dictionary is
	* structured.

Describe the solution you'd like

A new implementation of UUID generator which can reduce the storage size for the field

Related component

Indexing:Performance

Describe alternatives you've considered

No response

Additional context

No response

[mgodwan]

[Indexing Triage Meeting 07/22]

Next step by @mgodwan to share a sample implementation with JMH micro-benchmarks.